coordination-bridge

What problem does it solve?

Provides a resilient HTTP fallback for coordinator-dependent workflows so agents and scripts can continue claims, handoffs, memory operations, and guardrail checks when the MCP transport or coordinator socket is unavailable. It prevents hard failures in multi-agent coordination by detecting availability, probing capabilities, and degrading operations to safe "skipped" responses when appropriate.

Core Features & Use Cases

• Transport Detection: Probe /health and capability routes, or detect an MCP server, to determine whether to use HTTP or MCP transport.
• Capability Probing & Operations: Probe and invoke locks, work queue claim/submit, handoff read/write, memory store/query, guardrails checks, feature registry and merge-queue interactions over HTTP with normalized responses.
• Security & Safety: Enforces an allowlist for coordinator hosts (localhost by default, supports wildcard entries), validates URLs to prevent SSRF, accepts API keys via environment, and returns clear skipped/error states for unreachable or unauthorized endpoints.
• Use Case: Use this bridge in CI/CD or cloud-deployed agents to continue safe handoffs and memory lookups when the primary MCP transport is not reachable.

Quick Start

Run the coordination_bridge helper to detect coordinator reachability and obtain capability flags for subsequent bridge operations.